One particular case of such a structure is a debondable structure, in which the separation interface is an interface along which bonding via molecular adhesion has been carried out.
The expression “bonding via molecular adhesion” denotes bonding via intimate contact of the surfaces of the two substrates, employing adhesion forces, mainly van der Waals' forces, and not using an adhesive layer.
Without wishing to be limiting, it may, however, be considered that a debondable structure may be used mainly in four different applications:                a) Bonding of a mechanical stiffener: It may be desirable to bond a mechanical stiffener to a weak substrate or thin layer in order to prevent damaging or breaking it during certain fabrication steps, then to be able to remove this mechanical stiffener when its presence is no longer needed.        b) Rectifying poor bonding: Debonding makes it possible to debond two substrates that might not have been correctly bonded a first time, then to rebond them after cleaning, in order to improve the profitability of a fabrication process and to avoid, for example, scrapping poorly bonded substrates.        c) Temporary protection: During certain steps of storing or transporting substrates, especially in plastic boxes, it may be useful to temporarily protect their surfaces, especially those intended to be used subsequently for the fabrication of electronic components, in order to avoid any risk of contamination. One simple solution comprises bonding two substrates so that the faces thereof to be protected are bonded, respectively, to one another, then debonding these two substrates during the final use thereof.        d) Double transfer of a layer: This comprises producing a reversible bonding interface between an active layer and a first support substrate (optionally made of an expensive material), then transferring this active layer to a second final substrate by debonding the reversible bonding interface.        
However, applications may also be found in which it is desired to separate a structure formed of two assembled substrates along an interface that is not a bonding interface.
Such an interface may be, for example, an interface between a first material and a second material, which may have been joined to one another by an addition of the second material to the first, for example, via deposition, epitaxy, etc.
As a variant, such an interface may be, for example, a weak zone formed within a material and marked by the presence of bubbles, inclusions, etc. Separation along an interface that is not a bonding interface may, in particular, find an application in the transfer of a layer from a first substrate to a second substrate.
The layer to be transferred may thus not have been formed by bonding to the first substrate but, for example, may be formed by epitaxy or deposition on the substrate, or, alternatively, may be part of a thicker layer within which it has been delimited by a layer of bubbles that weakens the thick layer.
Irrespective of the envisaged applications, it is necessary to carry out the separation without damaging, scratching or contaminating the surface of the two substrates located on either side of the separation interface and without breaking these two substrates.
Depending on the various applications, these two “substrates to be separated” may be two layers of one and the same substrate or two different substrates.
Moreover, the larger the dimensions of the two substrates of the structure to be separated or the higher their bond energy, the more difficult the separation is to carry out, in particular, without damage.
Furthermore, it is known from the research studies by Maszara regarding the measurement of the bonding energy between two substrates (see the article by W. P. Maszara, G. Goetz, A. Caviglia and J. B. McKitterick: J. Appl. Phys. (1988), 64:4943) that it is possible to measure the bonding energy between two substrates by introducing a thin blade between the two at their bonding interface.
Maszara established the following relationship:
  L  =                    3        ⁢                                  ⁢                  Et          3                ⁢                  d          2                            32        ⁢                                  ⁢        γ              4  in which d represents the thickness of the blade inserted between the two bonded substrates, t represents the thickness of each of the two bonded substrates, E represents the Young's modulus along the debonding axis, γ represents the bonding energy and L represents the length of the crack between the two substrates at equilibrium.
The above formula starts from the hypothesis that the two substrates are of identical dimensions.
Owing to the aforementioned relationship, it is possible, by measuring L, to determine the bonding energy γ.
This definition of the “bonding” energy is based on the hypothesis that the energy needed to separate the two substrates, or rupture energy of the interface (which is the energy actually measured by the method using a blade), is equal to the bonding energy of the substrates.
In reality, during the separation of the substrates, a portion of the energy is dissipated, not in the rupture of the interface itself, but in other phenomena, such as deformations of the material(s) present at the interface.
In the remainder of the text, the rupture energy of an interface will, therefore, denote the energy to be provided in order to separate two substrates or layers along the interface.
Insofar as these substrates or layers to be separated are stiff enough to be separated with a blade, it is possible to separate them by parting them sufficiently from one another at their beveled edge, which has the effect of creating a separation wave. This wave propagates from the point of the edge of the substrates where it is initiated, across the entire surface of these substrates.
Furthermore, the separation is known to be assisted by the phenomenon referred to as stress corrosion.
Stress corrosion comprises, in combination with the parting force of the blade, the application of a fluid to the separation interface.
Stress corrosion is particularly beneficial when at least one of the substrates is made of silicon and when the interface comprises silicon oxide, whether it is a native oxide or an oxide formed intentionally, for example, in order to form a bonding layer or an insulating layer.
This is because such an interface contains siloxane (Si—O—Si) bonds that are broken by water molecules provided by the fluid.
The rupture energy of the interface is thus significantly reduced.
For the description of the stress corrosion process, reference may be made to chapter 14, entitled “Debonding of Wafer-Bonded Interfaces for Handling and Transfer Applications” by J. Bagdahn and M. Petzold, in Wafer Bonding: Applications and Technology, edited by M. Alexe and U. Gösele, Springer, 2004.
In particular, Cha et al., in “Why debonding is useful in SOI?,” Electrochemical Society Proceedings, Vol. 99-35, pp. 119-128, propose a two-step separation comprising partially parting the substrates by means of a blade, then introducing deionized water into the space thus formed until the complete separation of the substrates is achieved.
However, in certain applications, the interface along which it is desired to carry out the separation has a very high rupture energy, for example, greater than 1 J/m2, or even greater than 1.5 J/m2.
Such is the case, for example, for a structure of silicon-on-insulator (SOI)-type, or more broadly, for a structure of semiconductor-on-insulator (SeOI)-type, which comprises a support substrate, a buried dielectric layer (for example, an oxide layer) and a semiconductor layer (of silicon in the case of an SOI).
When this structure is produced by layer transfer, that is to say, by assembling a donor substrate comprising the semiconductor layer and a support substrate, the dielectric layer being at the interface, a heat treatment is generally carried out that aims to increase the rupture energy at the interface.
This makes it possible to prevent the structure from separating during the transfer of the semiconductor layer during subsequent SOI treatment steps.
By way of example, it is thus possible to attain a rupture energy of the order of 1.6 J/m2 at the interface.
However, due to this very high rupture energy, if an attempt is made to insert a blade in order to separate the semiconductor layer, there is a high risk of breaking the layer instead of separating it along the interface.
U.S. Pat. No. 7,713,369 proposes a process for fabrication of a detachable structure consisting of the assembly of two substrates, in which a peripheral zone having a high rupture energy and a central zone having a low rupture energy are formed at the interface, here, a bonding interface.
Thus, in order to separate the two substrates, the peripheral zone is removed by chemical etching or with the aid of a laser, until the central zone is reached, at which time it is possible to carry out a mechanical separation (for example, using a pressurized water or air jet, by pulling or by insertion of a blade).
However, the formation of these two zones of differing rupture energies on one and the same interface is restrictive to implement.
Furthermore, certain structures may comprise several interfaces, the separation of the substrates necessarily taking place at the interface that has the lowest rupture energy. This interface, predefined by its technical features, may not be that which a user would have chosen in order to carry out the separation of the substrates at the desired location.
One objective of the disclosure is, therefore, to propose a separation process that makes it possible to separate two assembled structures, along one interface chosen from among others, the interfaces having very high bond energies, which are identical or different, without the risk of fracturing or damaging the substrates.